Colicin E2 Is a DNA Endonuclease

Proc. Nati. Acad. Sci. USA Vol. 73, No. 11, pp. 3989-S993'November 1976 Biochemistry Colicin E2 is a DNA endonuclease (colicin-E2-immunity protein/colicin E3/colicin-E3-immunity protein) KLAUS SCHALLER AND MASAYASU NOMURA University of Wisconsin, Institute for Enzyme Research, 1710 University Avenue, Madison, Wisc. 53706 Communicated by Henry Lardy, September 10, 1976

ABSTRACT Colicin E2 purified by conventional methods immunity protein (20). These authors demonstrated that the contains a tightly bound low-molecular-weight protein, as has immunity protein could be separated from the E3 protein by been found with purified colicin E3 [Jakes, N. & Zinder, N. D. preparative electrophoresis in sodium dodecyl sulfate/poly- (1974) Proc. Nat. Acad. Sci. USA 71, 3380-33841. Such E2 preparations do not cause DNA cleavage in vitro. After sepa- acrylamide gels and that E3 protein (called "ES*") free of ration from the low-molecular-weight protein, colicin E2 re- immunity protein was much more active than the "complexed" tained the original in vivo killing activity, and in addition ES preparation in ribosome inactivation in vitro (20). They also showed a high activity in vitro in cleaving various DNA mole- noted the presence of a small-molecular-weight protein in cules, such as a ColE1 hybrid plasmid and DNAs from Esche- highly purified E2 preparations (cited in ref. 20). From the richia coli, X phage, 4X174 phage, and simian virus 40. The low-molecular-weight protein ("E2-immunity protein") specif- analogy of the colicin E3-immunity protein complex, they ically prevented this in vitro DNA cleavage reaction, i.e., had inferred that this small-molecular-weight protein is probably an "immunity function." The results demonstrate that colicin the E2-immunity protein. We have now succeeded in sepa- E2 itself is a DNA endonuclease and explain the in vivo effects rating colicin E2 (called "E2*") from the low-molecular-weight caused by E2 in sensitive cells as well as the mechanism of im- protein and demonstrated that E2* thus obtained is highly munity in E2-colicinogenic cells. active in degrading DNA in vitro and that the small-molecu- Colicin E2 (E2) is a protein antibiotic synthesized by certain lar-weight protein prevents this DNase activity, i.e., has an strains of coliform bacteria that carry the ColE2 plasmid, and "immunity function." is classified in the E group of colicins along with colicins El and E3 (1, 2). Although the physical properties of E2 and E3 are MATERIALS AND METHODS similar (3) and both colicins share the same receptor (1, 2), their E. coli PR13, which lacks RNase I, was used as the colicin- apparent mode of action is different. It was first discovered that sensitive strain. Colicins E3 and E2 were purified from mi- E2 causes degradation of DNA, while colicin E3 (E3) causes tomycin-C-induced cultures of E. coli CA38 and W3110 specific inhibition of protein synthesis in sensitive Escherichia (ColE2), respectively, according to the method described by coli cells treated with these colicins (4). Subsequent work on the Herschman and Helinski (3). The colicins obtained were still mode of action of E3 has demonstrated that ES inactivates ri- complexed with immunity proteins (cf. ref. 20). E3-immunity bosomes in treated cells (5) by causing the cleavage of a frag- protein was previously purified by J. Sidikaro in this laboratory ment from the 3' end of the 16S RNA (6, 7). The same cleavage (21). Pure E2-immunity protein was obtained as described reaction has also been demonstrated in vitro using "purified" below. E3 protein (see below) and purified ribosomes (8, 9). This in Colicin E2* was separated from the immunity protein by vitro demonstration of E3 action has strongly indicated that E3 dissociation of the complexed E2 with guanidine-HCl at about acts directly on ribosomes rather than indirectly and that E3 4°. Lyophilized E2 was dissolved in dissociation buffer (1 itself is an RNase with a very stringent substrate specificity. mg/ml). The dissociation buffer contained 6 M guanidine-HCl In contrast to ES, the mechanism of action of E2 has not been (Grade I, Sigma Chemical Co.), 0.2 M NaCl, 1 mM di- clear. Since the initial discovery of DNA degradation in E2- thiothreitol, and sufficient 1 M K2HPO4 to adjust the pH to 6.8. treated sensitive cells (4), several authors have confirmed and After the incubation for 1 hr at 40, the solution was subjected extended the original observations on DNA cleavage reactions to one of two methods to obtain a complete separation of E2* (10-13). However, others have suggested that the primary ac- from the immunity protein. In the first method, the solution tion of E2 is not on the cellular DNA, but on some other targets was applied to a Sephadex G-100 "superfine" column equili- such as membranes (14) or tRNA (15). brated with the dissociation buffer and the two proteins were Since E2 has properties similar to E3 (3), and E3 causes separated. In the second method, the solution was filtered ribosome inactivation in vitro, one might expect some kind of through a PM30 membrane (Amicon) using a 50 ml stirred cell. effects of E2 on cellular DNA in vitro, if DNA is in fact the The protein left in the Amicon cell (E2*) was diluted with the primary target of E2. Several attempts have been made to dissociation buffer and the filtration process was repeated. After demonstrate such effects in vitro using purified E2 (e.g., refs. five such processes, the E2* preparation was diluted in a solu- 16-19). However, published results were either negative or too tion of 6 M urea and 0.2 M NaCl, concentrated again to about uncertain to allow a definitive conclusion regarding such direct 2 mg/ml of protein, and dialyzed against 0.1 M sodium phos- effects. phate, pH 7.0, containing 0.2 M NaCI. The filtrate containing Recently, Jakes and Zinder observed that the "purified" ES the E2-immunity protein was concentrated using a UM2 filter preparations obtained with the conventional purification (Amicon) and treated in the same way as E2*. The proteins methods (3) contained about one molar equivalent of the E3- could be stored in the dialysis buffer at 40 in polypropylene tubes for at least 2 months without loss of in vivo or in vitro Abbreviations: E2 and ES, colicins E2 and ES, respectively; Col, a activity. Colicin ES* was prepared in a similar way. colicinogenic plasmid; kb, kilobase (1000 base pairs); SV40, simian virus Colicin activity in vivo was determined by spot-testing 40. dilutions of colicin in 10 mM potassium phosphate containing 3989 Downloaded by guest on October 2, 2021 3990 Biochemistry: Schaller and Nomura Proc. Natl. Acad. Sci. USA 73 (1976) Table 1. In vivo killing activity of colicins described in Materials and Methods. Jakes and Zinder were unable to recover active E3* after 6 M guanidine.HCl treat- E3 E3* E2 E2* ment and hence used sodium dodecyl sulfate to obtain E3* (20). ± 1 ± 0.5 1 ± 0.5 Using E3* prepared by our method (compare Fig. lb), we have 1 0.5 1 0.5 confirmed the observation made by Jakes and Zinder (20) that Values given are activity units per ng of colicin as defined by E3* is much more active in vitro (about 40-fold) than the Helinski (3). original E3 preparation. In addition, we have observed that the killing activity of E3* was similar to that of the original E3 2 mg/ml of bovine serum albumin on an agar layer of sensitive preparation when tested in vivo by the spot-test method (Table cells (3). To follow the in vitro DNA cleavage reaction, E2* was 1). In contrast to our results, Jakes.and Zinder noted that their incubated with DNA (usually covalently closed circular pLC E3* is 50-500 times less active in vivo than the original E3 (20). 21-9 plasmid DNA) at 370, and samples were analyzed using We conclude that the difference between our results and their slab gel electrophoresis with 0.9% agarose (22). The incubation results is probably due to the difference in the method of mixture contained 20mM Tris-HCI, pH 8.0,80mM NaCl, and preparation of E3* and that the E3-immunity protein is not 10 mM MgSO4. required for killing of intact cells by E3* (see Discussion). pLC 21-9 plasmid DNA was prepared by M. Kenerley in this laboratory from a derivative of E. coli strain MV12 (F+, recA -, Degradation of DNA by E2* thr-, leu-, thi-, Atrp E5) which carries this plasmid. The plasmid was originally constructed by Clarke and Carbon by We have applied the guanidine.HCl dissociation method to connecting an E. coli DNA fragment to ColEl DNA (23). The prepare E2* free of the presumed E2-immunity protein. Fig. method used to prepare this plasmid DNA was similar to that 1 shows the results of polyacrylamide/urea gel analysis of the used to make ColE3 DNA (21). Electron microscopic studies original E2 preparation (gel d), E2* (gel e), and the presumed showed that the plasmid DNA is 12,400 base pairs (12.4 kb) in immunity protein (gel f). We have found that E2* thus pre- length and consists of ColEl DNA (6.5 kb) and bacterial DNA pared degrades various DNAs in vitro as analyzed by agarose (M. Kenerley and M. Nomura, unpublished experiments). The gel electrophoresis. In addition, we also demonstrated that the DNA was mostly in the covalently closed circular form (see, e.g., presumed E2-immunity protein has in fact the ability to inhibit Figs. 2 and 4). E. coli DNA (24) and X phage DNA (25) were the DNase activity of E2*. prepared as described previously. Both simian virus 40 (SV40) In the experiment shown in Fig. 2, pLC 21-9 plasmid DNA DNA and hamster liver DNA were gifts from B. Weisblum, and (mostly in the form of covalently closed circular DNA) was phage 4X174 DNA was a gift from R. D. Wells. Phage R17 incubated with various amounts of E2* for 1 hr at 370 and re- RNA was prepared according to (26). action mixtures were analyzed by gel electrophoresis. It can be seen that at lower concentrations of E2* (gels a to e compared RESULTS with the control, gel m), conversion of covalently closed DNA to open circles (and possibly linear "whole" molecules) is ap- Preparation of colicin E3* free of the E3-immunity parent (see also below). With 0.25 ,ug of E2* (gel e), it is clear protein using guanidine-hydrochloride that this conversion is almost complete, and yet no significant We originally developed a method to dissociate the E3-im- production of smaller DNA fragments was observed. This in- munity complex and to recover active E3*. The method indicates that under these conditions E2* probably caused mainly volves dissociation of the complex by 6 M guanidine-HCl as single-strand DNA scissions. At higher E2* concentrations, the amount of open circular DNA of high molecular weight de- abc de f creased and the production of smaller DNA fragments with random size distribution was observed (gels f-i). *1_ a b c de f gh i j k I m oc\ WL CCC//

FIG. 2. Degradation of pLC 21-9 plasmid DNA by E2* in vitro. Plasmid DNA (10 Mg) was incubated with various amounts of E2* in a total volume of.100 Mul for 1 hr at 370. Twenty-five microliters from FIG. 1. Electrophoresis of colicin E3 and E2. Purified colicin E3 each reaction mixture was analyzed by agarose gel electrophoresis. and E2 were electrophoresed on 10% polyacrylamide disc gels con- The different amounts of colicin E2* (in Mg) were 0.012 (a), 0.025 (b), taining 0.1% sodium dodecyl sulfate and 6 M urea in 0.1 M sodium 0.05 (c), 0.12 (d), 0.25 (e), 0.5 (f), 1.25 (g), 2.5 (h), 3.25 (i); 2.5 Mug of E2* phosphate, pH 7.2 (32). The samples were E3 (a), E3* (b), E3-im- preincubated with 2.5 Mg of E2-immunity protein for 30 min at 370 munity protein (c), E2 (d), E2* (e), and E2-immunity protein (f). In (j); 2.5 Mg of E2-immunity protein without E2* (k), 10g of complexed the preparation shown in gel (d), we observed, in addition to E2* and E2 (1); and no addition, i.e., of untreated plasmid DNA (m). Three the E2 immunityprotein, two protein bands that presumably repre- arrows indicate the positions of open circular (OC), whole linear (WL), sent impurities. During the subsequent purification process these and covalently closed circular (CCC) plasmid DNA molecules. For "impurities" were apparently removed. WL, see also Fig. 4. Downloaded by guest on October 2, 2021 Biochemistry: Schaller and Nomura Proc. Natl. Acad. Sci. USA 73 (1976) 3991

b c d e f g h i i a b c de f g h i i

WL ccc

FIG. 4. Effect of E2-immunity protein on the second stage of E2*-induced DNA degradation in vitro. A 300 ,l reaction mixture FIG. 3. Specific inhibition of the E2*-induced in vitro DNA containing 34 ,g of pLC 21-9 plasmid DNA and 0.07 ztg of E2* was cleavage reaction by E2-immunity protein. E2* (1 Mg) was preincu- incubated at 370 (molar ratio of DNA to E2* 3:1). After 1 hr 100 gul bated for 30 min at 37° with various amounts of E3-immunity protein were removed and incubated separately with 0.05 ,gg of E2-immunity [5.0 ,ug (e), 2.5 Mg (f)] or E2-immunity protein [0.5 Mg (g), 0.25 ,ug (h)] protein. The remainder was incubated without immunity protein. in 40 Ml of reaction mixture. pLC 21-9 plasmid DNA (6.5 Mg in 10 Ml) Samples (30 Mul each) were taken at different timepoints from both was then added, and incubation was continued for one additional mixtures and cooled to 00. A control with DNA only is shown in (a). hour. Samples were then analyzed as described in the legend to Fig. Portions were taken from the reaction mixture without E2-immunity 2. Controls contained DNA only (a), E2* without immunity protein protein at "zero time" (b), at 1 hr incubation (c), 3 hr (e), 6 hr (g), and (b), 0.5 Mg of E3-immunity protein (c), and 0.5 Mg of E2-immunity 12 hr (i). From the reaction mixture with E2-immunity protein added protein (d). In a separate experiment the plasmid DNA was incubated at 1 hr, samples were taken at 3 hr (f), 6 hr (h), and 12 hr (j). X DNA with 1 Mg of E3* (j) and without (i), and DNAs were analyzed in a digested with HindIII restriction endonuclease provided the size similar way. The results shown here indicate that the amount of standards (d). The sizes of the fragments (33) are given in kilobases E2-immunity protein needed to titrate E2* is between 0.25 (gel h) and (kb). OC, WL, and CCC are defined in the legend to Fig. 2. In this 0.5 ,g (gel g) per 1 Mg of E2*. From the molecular weights ofE2* and experiment its position indicates that WL is about 14 kb in size. Using E2-immunity protein, this corresponds to about 1.2 to 2.4 immunity other DNA fragments as standards, the size of WL was estimated to protein molecules per E2* molecule. Further studies are required to be about 12.5 kb. These values agree with 12.4 ± 0.7 kb length ofthe determine the exact stoichiometry for the inhibition of E2* by the intact plasmid DNA determined by electron microscopic measure- immunity protein. ments. In contrast to E2*, the original "purified" E2 preparation in vitro (data not shown). We conclude that the observed in failed to cause any significant nucleolytic activity on the plas- vitro DNA cleavage reaction is due to the E2* protein itself and mid DNA (gel 1). This can be explained by the presence of the not due to contaminating nonspecific nucleases such as endo- presumed E2-immunity protein in the preparation, since the nuclease I. latter protein inhibited the nucleolytic activity of E2* com- As described above (compare Fig. 2; and also experiments pletely (gel j compared to gel h). We call this protein "E2- shown in Fig. 4), the first event in the in vitro degradation of immunity protein." The E2-immunity protein itself has no pLC 21-9 plasmid DNA by E2* appears to be a single-strand significant DNase activity (compare gel k). scission to convert the double-stranded covalently closed circular DNA to an open circle. The first reaction ("the first stage") Specificity and general features of in vitro DNA is followed by subsequent DNA cleavage ("the second stage") cleavage reaction by E2* resulting in the production of linear DNAs and gradual de- creases in the average size of the DNA (Fig. 4; see also Fig. 2). As described above, plasmid DNA is hydrolyzed by E2*. The cleavage reaction in this second stage is also due to the E2* However, E3*, which is highly active in ribosomal RNA protein itself, since addition of E2-immunity protein cleavage, has no such nucleolytic activity on pLC 21-9 DNA during (Fig. 3, gels i and j). Both E2* and E3* appear to be pure (see a bc abc a b c abc abc Fig. lb and e). In addition, since the method of preparation of E2* is almost identical to that of E3*, it seems unlikely that the DNA cleavage reaction induced by E2* in vitro is due to some nucleases such as endonuclease I contaminating E2* preparations. Moreover, E3-immunity protein, which has the ability to inhibit the ribosome inactivation induced by E3 in nitro (21, 27), is unable to inhibit the DNA cleavage reaction induced by E2 in vitro even at an amount 10 times higher than that ef- fective with E2-immunity protein (Fig. 3, gels e and f compared to gels b, g, and h). We have also shown that partially purified endonuclease I degrades the plasmid DNA, and that the E2- immunity protein failed to inhibit this degradation (data not E coli A SV40 *X174 R17 shown). Therefore, the specific inhibition of the E2*-induced FIG. 5. Action of colicin E2* on different nucleic acids. E2* (1 DNA cleavage reaction by the purified E2-immunity protein Mug) was preincubated for 30 min at 37° without (a) and with (b) 0.5 also argues against the possibility that endonuclease I contam- Mg of E2-immunity protein. Various nucleic acid preparations (as ination of E2* preparations is responsible for the observed DNA indicated in the figure) were then added and incubated for one additional hour. Sample (c) was an incubation of the nucleic acid alone. cleavage reaction. Transfer RNA, which is known to inhibit The amounts ofsubstrate in 50 Ml reaction mixture were 2.65 Mg ofE. endonuclease I (ref. 28; and our own unpublished experiments), coli DNA, 5,ug of X DNA, 1 Mg of SV40 DNA, 5 ,ug of kX174 DNA, and also failed to inhibit the DNA cleavage reaction induced by E2* 5 Mg of R17 RNA. Samples were analyzed as before. Downloaded by guest on October 2, 2021 3992 Biochemistry: Schaller and Nomura Proc. Natl. Acad. Sci. USA 73 (1976) the second stage prevents further DNA cleavage (Fig. 4, gels one involving an activation of cellular endonuclease I by E2 f, h, and j compared to gels e, g, and i, respectively). (15). Catalytic nature of the DNA cleavage reaction induced Saxe (19) previously reported a very weak endonucleolytic by E2* activity of E2 on supercoiled X DNA in vitro (less than o-4 times the activity shown in our experiments). The possibility In the experiment described in Fig. 4, the molar ratio of E2* could not be excluded that the observed nucleolytic activity was to DNA was 1 to 3. In this experiment, all the DNA molecules due to some nuclease copurified with E2 [see the Discussion in underwent at least two and probably several cleavage events the paper by Saxe (19)]. Since Saxe used a colicin E2 preparation caused by E2*. [The size of the original plasmid DNA was 12.4 which was complexed with the E2-immunity protein, the weak kb. The degradation products showed broad size distribution activity could be explained by the inhibitory action of the from 10 kb to less than 2 kb (Fig. 4, gel i compared to the ref- E2-immunity protein on the nuclease activity of E2* as dem- erence DNA fragments, gel d).] Moreover, because of the pos- onstrated in this paper. In fact, we failed to detect any nu- sibility that many inactive molecules could be present in our cleolytic activity with purified E2 complexed with the immu- E2* preparation, the actual number of cleavages by one active nity protein (see Fig. 2, gel 1). E2* molecule might be much higher than that indicated in our Earlier in vivo studies by Ringrose (11) showed that there are experimental results. Therefore, it is apparent that one molecule three distinguishable stages in the degradation of DNA in E2* is able to cleave DNA several, and probably many, times, E2-treated E. coli cells: stage I, corresponding to single-strand hence colicin E2* acts in a catalytic way like an enzyme. DNA cleavage; stage II, double-strand DNA breakage; and Substrate specificity of the DNA cleavage reaction in stage III, formation of acid-soluble DNA breakdown products. vitro Our in vitro experiments indicate that DNA breakdown corresponding to both stages I and II involves the direct action of The above experiments on the E2*-induced cleavage in vitro E2* on DNA, but that cellular nuclease(s), and not colicin E2, were done mainly using pLC 21-9 plasmid DNA as a "sub- is responsible for stage III. strate. " We have also demonstrated that E2* can degrade the The present studies have shown that the basic mechanism following DNAs in vitro: E. coli DNA, XfusS transducing phage of immunity to E2 in E2-colicinogenic cells is analogous to the DNA, XDNA, SV40 DNA, hamster liver DNA, and 4X174 mechanism elucidated for the colicin E3 system and discussed DNA. In every case, DNA degradation was inhibited by E2- in the previous papers (9, 21, 27). The E2-immunity protein immunity protein (Fig. 5; and other data not shown). In contrast used in the present studies was previously isolated by Jakes and to these DNAs, RNA prepared from phage R17 was not de- Zinder (cited in ref. 20) and its role in E2 immunity was hy- graded (Fig. 5). We conclude that E2* can degrade a variety pothesized. This hypothesis has now been proven experimen- of DNAs, both double-stranded and single-stranded DNAs as tally. well as both prokaryotic and eukaryotic DNAs, but not RNA. Jakes and Zinder observed that the in vivo killing activity of Exact nucleotide sequence specificity of the cleavage sites has the E3* prepared by them was much weaker than that of E3 not been studied. We note that the apparent broad substrate complexed with the E3-immunity protein, and considered the specificity of the E2*-induced DNA cleavage reaction is in possibility that E3-immunity protein may have a role in the marked contrast to the more stringent substrate specificity attachment of E3 to receptors on sensitive cells (20). However, observed in the in vitro RNA cleavage reaction of E3*. as described in this paper, our experiments show that E3- (and is not for the in Absence of DNA exonuclease activity in E2* E2-) immunity protein required vivo killing preparations action of E3 (and E2), and hence for the attachment of the colicins to their receptors. The binding of E3* to isolated E3- The results described above demonstrate that E2* has a DNA receptors was also demonstrated (our unpublished experi- endonuclease activity. However, E2* does not appear to have ments). any exonuclease activity. For example, we incubated radioac- Finally we note that our in vitro system for E2* is simpler tive E. coli DNA with E2* and found that no significant amount than the in vitro system for E3 studied previously. In the case of acid-soluble radioactive oligonucleotides was produced even of in vitro inactivation of ribosomes by E3, both 30S and 50S after a long period of incubation, when endonucleolytic clea- subunits are required, and free 16S rRNA cannot serve as a vages of the DNAs were extensive (data not shown). substrate (9, 30, 31), even though the reaction is a cleavage of 16S rRNA. Thus, one cannot completely exclude the possibility that ES itself is not the enzyme, but activates a latent ribonu- DISCUSSION clease (e.g., a ribosomal protein) contained in the ribosomes. Experimental results described in this paper demonstrate that Our E2* system consists of protein-free DNA and purified E2* purified colicin E2* free of E2-immunity protein has a DNA protein, and we have been able to conclude that E2* itself has endonuclease activity and that the E2-immunity protein spe- a nuclease activity. Thus, we believe that the possibility that ES* cifically prevents this activity. These in vitro results can account activates a ribosomal RNAse is extremely unlikely and that ES* for the in vivo biochemical effects of E2 on sensitive cells, and protein itself has a nucleolytic activity that results in the 16S the phenomenon of the immunity in E2-colicinogenic cells. rRNA cleavage. Although various biochemical and physiological changes in addition to the DNA degradation were reported in E2-treated We thank Dr. D. L. Nelson and L. Post for reading the manuscript. cells (see introduction), none of these changes take place in the This work was supported in part by the College of Agriculture and Life E2-colicinogenic cells. Therefore, we conclude that the primary Sciences, University of Wisconsin, and by grants from the United States target of colicin E2 is DNA. Any other changes observed, such National Science Foundation (GB-31086) and the National Institutes as slow degradation of RNA (29) or inhibition of cell division of Health (GM-20427). K.S. is a recipient of a postdoctoral fellowship (14), must be secondary consequences of the DNA cleavage in from the Deutsche Forschungsgemeinschaft. E2-treated cells. Our experiments also exclude several suggested mechanisms for the indirect action of E2 on DNA, such as the 1. Fredericq, P. (1958) Symp. Exp. Biol. 12, 104-122. Downloaded by guest on October 2, 2021 Biochemistry: Schaller and Nomura Proc. Natl. Acad. Sci. USA 73 (1976) 3993

2. Nomura, M. (1967) Annu. Rev. Microbiol. 21, 257-284. 18. Maeda, A. (1974) Biochim. Blophys. Acta 374,426-430. 3. Herschman, H. R. & Helinski, D. R. (1967) J. Blol. Chem. 242, 19. Saxe, L. S. (1975) Biochemistry 14,2058-2063. 5360-5368. 20. Jakes, K. & Zinder, N. D. (1974) Proc. Natl. Acad. Sci. USA 71, 4. Nomura, M. (1963) Cold Spring Harbor Symp. Quant. Biol. 28, 3380-3384. 315-324. 21. Sidikaro, J. & Nomura, M. (1975) J. Biol. Chem. 250, 1123- 5. Konisky, J. & Nomura, M. (1967) J. Mol. Btol. 26,181-195. 1131. 6. Bowman, C. M., Dahlberg, J. E., Ikemura, T., Konisky, J. & No- 22. Shinnick, T. M., Lund, E., Smithies, 0. & Blattner, F. R. (1975) mura, M. (1971) Proc. Natl. Acad. Sci. USA 68,964-968. Nucleic Acids Res. 2, 1911-1929. 7. Senior, B. W. & Holland, I. B. (1971) Proc. Nati. Acad. Sd. USA 23. Clarke, L. & Carbon, J. (1975) Proc. Natl. Acad. Sci. USA 72, 68,959-963. 4361-4365. 8. Boon, T. (1971) Proc. Nati. Acad. Sci. USA 68,2421-2425. 24. Marmur, J. (1961) J. Mol. Biol. 3,208-218. 9. Bowman, C. M., Sidikaro, J. & Nomura, M. (1971) Nature New 25. Zabay, G., Chambers, D. A. & Cheong, L. C. (1970) in The Btol. 234, 133-137. Lactose Operon, eds. Beckwith, J. R. & Zipser, D. (Cold Spring 10. Obinata, M. & Mizuno, D. (1970) Biochim. Biophys. Acta 199, Harbor Laboratory, Cold Spring Harbor, N.Y.), pp. 375-391. 330-39. 26. Gesteland, R. F. & Boedtker, H. (1964) J. Mol. Biol. 8, 496- 11. Ringrose, P. (1970) Biochim. Biophys. Acta 213,320-34. 507. 12. Almendinger, R. & Hager, L. P. (1972) Nature New Biol. 235, 27. Jakes, K., Zinder, N. D. & Boon, T. (1974) J. Btol. Chem. 249, 199-203. 438-444. 13. Saxe, L. S. (1975) Biochemistry 14, 2051-2057. 28. Lehman, L. R., Roussos, G. G. & Pratt, E. A. (1962) J. Btol. Chem. 14. Beppu, T., Kawabata, K. & Arima, K. (1972) J. Bacteriol. 110, 237,819-828. 485-493. 29. Nose, K., Mizuno, D. & Ozeki, H. (1966) Btochim. Blophys. Acta 15. Almendinger, R. & Hager, L. P. (1973) in Chemistry and 119,636-8. Functions of Colicins, ed. Hager, L. P. (Academic Press, New 30. Boon, T. (1972) Proc. Natl. Acad. Sct. USA 69,549-552. York), pp. 107-133. 31. Bowman, C. M. (1972) FEBS Lett. 22,73-75. 16. Ringrose, P. S. (1972) FEBS Lett. 23, 241-243. 32. Weber, K. & Osborn, M. (1969) J. Btol. Chem. 244, 4406- 17. Seto, A., Shinozawa, T. & Maeda, A. (1973) Biochim. Biophys. 4412. Acta 324, 305-308. 33. Murray, K. & Murray, N. E. (1975) J. Mol. Biol. 98,551-564. Downloaded by guest on October 2, 2021